Safe or Stranded On Base: Part I
The main goal for a pitcher to get batters out. And if a batter reaches base, then the goal becomes stranding the runner on base. A lot of things change for a pitcher when there is a runner on base compared to having nobody on. His entire delivery to the plate changes, his attention cannot focus solely on the batter, and there is an increased amount of pressure. Whether clutch hitting exists is debatable, but for pitchers there are certainly pitchers that have more skill than others in getting batters out with runners on base. I know from my personal pitching experience that pitching from the stretch is incredibly different than pitching from the windup, and checking the runner forces you to start your delivery without first getting settled in and focused on the strike zone. Note that I'm saying some pitchers are better at pitching with runners on base, not that some pitchers are more clutch than others. There is a difference.
In the baseball statistics world, the rate at which pitchers strand runners on base is referred to as strand rate or Left On Base Percentage, LOB%. The league average LOB% from 05-07 was 71.96%, so basically 72%. Any pitcher with an LOB% either over or under the 72% can be considered lucky or unlucky, sort of. If you think about it, a better pitcher should have a higher LOB% because a better pitcher gives up less hits, right? Determining how much to deviate from 72% is not an exact science, but my method is using Batting Average Against vs. LOB% and then finding the slope of the best fit line.
This graph shows pitchers' Batting Average Againsts versus their LOB% from the 2005 to 2007 season. Notice that I inverted the bottom axis so the LOB percentages go up, but the slope of the line is still negative meaning LOB% and BAA are inversely related. As a general rule, for every .005 of BAA added, take off .5 from the LOB%. First you should know that for the league-average LOB% of 72%, the expected BAA is .265. So for example, if a pitchers BAA is .275, then their LOB% should be 71%. Remember however, with the R-Squared only at .29 this is more a guideline than anything else. Don't follow this rule religiously.
If you are wondering why I chose BAA instead of other stats like ERA, FIP, and WHIP, all of these stats did not work for various reasons. ERA did correlate especially well with LOB% (.573 R-Squared) but I felt Era was too dependant on LOB% and was not a good predictor of it. FIP correlated surprisingly low, .092 correlation, so I decided not to use it. Lastly, WHIP is similar to BAA, yet just did not correlate as well (.239 R-Squared). My preference would have been to use OPS-against, but those numbers are not so easily available, we can only use what we have.
Overall LOB% is definitely a stat that is overlooked far to often when evaluating pitchers. However, it is tricky to use since not every pitcher should be expected to regress to the average. Until a better method is found, (maybe by me) my +.005/-.5 rule is more than adequate for determining xLOB%.
This is only part one of two of my series on strand rate, next time we'll examine individual pitchers' splits with nobody on/runners on base and see if any conclusions can be drawn from those.
In the baseball statistics world, the rate at which pitchers strand runners on base is referred to as strand rate or Left On Base Percentage, LOB%. The league average LOB% from 05-07 was 71.96%, so basically 72%. Any pitcher with an LOB% either over or under the 72% can be considered lucky or unlucky, sort of. If you think about it, a better pitcher should have a higher LOB% because a better pitcher gives up less hits, right? Determining how much to deviate from 72% is not an exact science, but my method is using Batting Average Against vs. LOB% and then finding the slope of the best fit line.
This graph shows pitchers' Batting Average Againsts versus their LOB% from the 2005 to 2007 season. Notice that I inverted the bottom axis so the LOB percentages go up, but the slope of the line is still negative meaning LOB% and BAA are inversely related. As a general rule, for every .005 of BAA added, take off .5 from the LOB%. First you should know that for the league-average LOB% of 72%, the expected BAA is .265. So for example, if a pitchers BAA is .275, then their LOB% should be 71%. Remember however, with the R-Squared only at .29 this is more a guideline than anything else. Don't follow this rule religiously.
If you are wondering why I chose BAA instead of other stats like ERA, FIP, and WHIP, all of these stats did not work for various reasons. ERA did correlate especially well with LOB% (.573 R-Squared) but I felt Era was too dependant on LOB% and was not a good predictor of it. FIP correlated surprisingly low, .092 correlation, so I decided not to use it. Lastly, WHIP is similar to BAA, yet just did not correlate as well (.239 R-Squared). My preference would have been to use OPS-against, but those numbers are not so easily available, we can only use what we have.
Overall LOB% is definitely a stat that is overlooked far to often when evaluating pitchers. However, it is tricky to use since not every pitcher should be expected to regress to the average. Until a better method is found, (maybe by me) my +.005/-.5 rule is more than adequate for determining xLOB%.
This is only part one of two of my series on strand rate, next time we'll examine individual pitchers' splits with nobody on/runners on base and see if any conclusions can be drawn from those.
